cryptospore / rpms / qemu-kvm

Forked from rpms/qemu-kvm 2 years ago
Clone
9ae3a8
From 3be810287878138e5f72568d1ba1160b5bad22f8 Mon Sep 17 00:00:00 2001
9ae3a8
From: "Daniel P. Berrange" <berrange@redhat.com>
9ae3a8
Date: Thu, 8 Feb 2018 17:50:32 +0100
9ae3a8
Subject: [PATCH 18/27] ui: track how much decoded data we consumed when doing
9ae3a8
 SASL encoding
9ae3a8
MIME-Version: 1.0
9ae3a8
Content-Type: text/plain; charset=UTF-8
9ae3a8
Content-Transfer-Encoding: 8bit
9ae3a8
9ae3a8
RH-Author: Daniel P. Berrange <berrange@redhat.com>
9ae3a8
Message-id: <20180208175041.5634-19-berrange@redhat.com>
9ae3a8
Patchwork-id: 78951
9ae3a8
O-Subject: [RHEL-7.5 qemu-kvm PATCH v1 18/27] ui: track how much decoded data we consumed when doing SASL encoding
9ae3a8
Bugzilla: 1527405
9ae3a8
RH-Acked-by: Laszlo Ersek <lersek@redhat.com>
9ae3a8
RH-Acked-by: Gerd Hoffmann <kraxel@redhat.com>
9ae3a8
RH-Acked-by: Miroslav Rezanina <mrezanin@redhat.com>
9ae3a8
9ae3a8
From: "Daniel P. Berrange" <berrange@redhat.com>
9ae3a8
9ae3a8
When we encode data for writing with SASL, we encode the entire pending output
9ae3a8
buffer. The subsequent write, however, may not be able to send the full encoded
9ae3a8
data in one go though, particularly with a slow network. So we delay setting the
9ae3a8
output buffer offset back to zero until all the SASL encoded data is sent.
9ae3a8
9ae3a8
Between encoding the data and completing sending of the SASL encoded data,
9ae3a8
however, more data might have been placed on the pending output buffer. So it
9ae3a8
is not valid to set offset back to zero. Instead we must keep track of how much
9ae3a8
data we consumed during encoding and subtract only that amount.
9ae3a8
9ae3a8
With the current bug we would be throwing away some pending data without having
9ae3a8
sent it at all. By sheer luck this did not previously cause any serious problem
9ae3a8
because appending data to the send buffer is always an atomic action, so we
9ae3a8
only ever throw away complete RFB protocol messages. In the case of frame buffer
9ae3a8
updates we'd catch up fairly quickly, so no obvious problem was visible.
9ae3a8
9ae3a8
RHEL-7 note: context difference in the last argument to
9ae3a8
vnc_client_io_error() due to downstream lacking commit 04d2529da27d
9ae3a8
("ui: convert VNC server to use QIOChannelSocket", 2015-12-18).
9ae3a8
9ae3a8
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
9ae3a8
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
9ae3a8
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
9ae3a8
Message-id: 20171218191228.31018-6-berrange@redhat.com
9ae3a8
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
9ae3a8
(cherry picked from commit 8f61f1c5a6bc06438a1172efa80bc7606594fa07)
9ae3a8
Signed-off-by: Miroslav Rezanina <mrezanin@redhat.com>
9ae3a8
---
9ae3a8
 ui/vnc-auth-sasl.c | 3 ++-
9ae3a8
 ui/vnc-auth-sasl.h | 1 +
9ae3a8
 2 files changed, 3 insertions(+), 1 deletion(-)
9ae3a8
9ae3a8
diff --git a/ui/vnc-auth-sasl.c b/ui/vnc-auth-sasl.c
9ae3a8
index f3ad75d..804b8e7 100644
9ae3a8
--- a/ui/vnc-auth-sasl.c
9ae3a8
+++ b/ui/vnc-auth-sasl.c
9ae3a8
@@ -64,6 +64,7 @@ long vnc_client_write_sasl(VncState *vs)
9ae3a8
         if (err != SASL_OK)
9ae3a8
             return vnc_client_io_error(vs, -1, EIO);
9ae3a8
 
9ae3a8
+        vs->sasl.encodedRawLength = vs->output.offset;
9ae3a8
         vs->sasl.encodedOffset = 0;
9ae3a8
     }
9ae3a8
 
9ae3a8
@@ -75,7 +76,7 @@ long vnc_client_write_sasl(VncState *vs)
9ae3a8
 
9ae3a8
     vs->sasl.encodedOffset += ret;
9ae3a8
     if (vs->sasl.encodedOffset == vs->sasl.encodedLength) {
9ae3a8
-        vs->output.offset = 0;
9ae3a8
+        vs->output.offset -= vs->sasl.encodedRawLength;
9ae3a8
         vs->sasl.encoded = NULL;
9ae3a8
         vs->sasl.encodedOffset = vs->sasl.encodedLength = 0;
9ae3a8
     }
9ae3a8
diff --git a/ui/vnc-auth-sasl.h b/ui/vnc-auth-sasl.h
9ae3a8
index 8091d68..4ec6fb5 100644
9ae3a8
--- a/ui/vnc-auth-sasl.h
9ae3a8
+++ b/ui/vnc-auth-sasl.h
9ae3a8
@@ -54,6 +54,7 @@ struct VncStateSASL {
9ae3a8
      */
9ae3a8
     const uint8_t *encoded;
9ae3a8
     unsigned int encodedLength;
9ae3a8
+    unsigned int encodedRawLength;
9ae3a8
     unsigned int encodedOffset;
9ae3a8
     char *username;
9ae3a8
     char *mechlist;
9ae3a8
-- 
9ae3a8
1.8.3.1
9ae3a8